NASA Technical Reports Server (NTRS) 20160001336: A Numerical Approach to Estimate the Ballistic Coefficient of Space Debris from TLE Orbital Data

1 MB·English

by  NASA Technical Reports Server (NTRS)

VILNIAUSUNIVERSITETAS
FIZIKOSFAKULTETAS
TEORINĖSFIZIKOSKATEDRA
JonasNarkeliūnas
ANUMERICALAPPROACHTOESTIMATETHEBALLISTICCOEFFICIENTOFSPACE
DEBRISFROMTLEORBITALDATA
Mokslotiriamasisdarbas
(studijųprograma–TEORINĖFIZIKAIRASTROFIZIKA)
Studentas JonasNarkeliūnas
Darbovadovas dr. JanStupl
Katedrosvedėjas
Vilnius2016
Contents
Introduction ................................................................................................ 3
1 AtmosphericDrag ..................................................................................... 5
1.1 DragForceandDragCoefficient........................................................... 5
1.2 AtmosphericDensityModel................................................................ 7
2 NumericalApproach .................................................................................. 10
2.1 InputData .................................................................................... 10
2.2 BallisticCoefficient.......................................................................... 12
2.3 IntegrationUsingSGP4..................................................................... 13
2.4 Fast-FourierTransformation ................................................................ 14
3 Results .................................................................................................. 18
Conclusion ................................................................................................. 24
References .................................................................................................. 25
2
Introduction
LowEarthOrbit(LEO)isfullofspacedebris,whichconsistofspentrocketstages,oldsatellitesand
fragments from explosions and collisions. As of 2009, more than 21 000 orbital debris larger than
10cmareknowntoexist[1],andwhileitishardtotrackanythingsmallerthanthat,theestimated
population of particles between 1 and 10 cm in diameter is approximately 500 000, whereas small
as1cmexceeds100million. TheseobjectsorbitEarthwithhugekineticenergies–speedsusually
exceed7km/s. Theshapeoftheirorbitvariesfromalmostcirculartohighlyellipticalandcoversall
LEO,aregioninspacebetween160and2000kmabovesealevel. Unfortunately,LEOisalsothe
placewheremostofouractivesatellitesaresituated1,aswellas,InternationalSpaceStation(ISS)
andHubbleSpaceTelescope,whoseorbitsarearound400and550kmabovesealevel,respectively.
This poses a real threat as debris can collide with satellites and deal substantial damage or even
destroythem.
Collisionsbetweentwoormoredebriscreatecloudsofsmallerdebris,whicharehardertotrack
andincreaseoverallobjectdensityandcollisionprobability. Atsomepoint,thedebrisdensitycould
thenreachacriticalvalue,whichwouldstartachainreactionandthenumberofspacedebriswould
grow exponentially. This phenomenon was first described by Kessler in 1978 and he concluded
that it would lead to creation of debris belt, which would vastly complicate satellite operations in
LEO.[2].
The debris density is already relatively high, as seen from several necessary debris avoidance
maneuvers donebyShuttle,beforeitwasdiscontinued,andISS[1,3,4]. Butnotallsatelliteshave
apropulsionsystemtoavoidcollision,hencedifferentmethodsneedtobeapplied. Oneofthepro-
posed collision avoidance concepts is called LightForce and it suggests using photon pressure to
inducesmallorbitalcorrectionstodeflectdebrisfromcolliding[5,6]. Thismethodisveryefficient
– as seen from theoretical simulations, even few continuous mode 10 kW ground-based lasers, fo-
cused by 1.5 m telescopes with adaptive optics, were enough to prevent significant amount of the
debriscollisions. SimulationsweredonebypropagatingallspaceobjectsinLEOby1yearintothe
futureandcheckingwhethertheprobabilityofcollisionwashigh. Forthosespaceobjectsdifferent
ground-based lasers were used to divert them, afterwards collision probabilities were reevaluated.
However,theactualaccuracyoftheLightForcesoftware,whichhasbeendevelopedatNASAAmes
ResearchCenter,dependsontheveracityoftheinputparameters,oneofwhichistheobject’sbal-
listiccoefficient. Itisameasureofbody’sabilitytoovercomeairresistance,whichhasasignificant
impact on the debris in LEO, and thus it is responsible for the shape of the trajectory of the de-
bris. Having the exact values of the ballistic coefficient would make significantly better collision
predictions,unfortunately,wedonotknowwhatarethevaluesformostoftheobjects.
In this research, we were working with part of LightForce code, which estimates the ballis-
1Source: https://www.space-track.org.
3
tic coefficient from ephemerides2. Previously used method gave highly inaccurate values, when
compared to known objects, and it needed to be changed. The goal of this work was to try out a
differentmethodofestimatingtheballisticcoefficientandtocheckwhetherornotitgivesnoticeable
improvementsinaccuracy. Herewepresentournewapproachandtheideasbehindit.
2Ephemeris(pluralephemerides)givesthepositionofastronomicalobjectsaswellasspacedebrisintheskyata
giventime.
4
1 Atmospheric Drag
Everything above Karman line, 100 km above sea level, is assumed to be outer space, however,
atmosphere extends far beyond that point. It extends for at least 10 000 km more [7], although
after 600 - 700 km it is not very significant (see Figure 1.1). Nevertheless, besides gravity, the
atmosphericdrageffecthasthemostimpactontheobject’strajectoryinLEOandithastobetaken
into account when modeling orbits of space objects in LEO. From here everything will eventually
fall down to Earth, but the magnitude of the effect depends on many different parameters and we
presenttheminthissection.
1.1 Drag Force and Drag Coefficient
An object moving through medium always experiences drag – a resistance force acting in the op-
posite direction of object’s movement. It is a result of three main effects: surface friction, flow
interference and form drag. At relatively high velocities and low densities, like in our case, the
lateral effect overshadows the first two, so they can be neglected [9]. In this case, the drag force is
largelyproportionaltocharacteristicareaoftheobjectandtothesquareofitsvelocity.
Figure1.1Variationsofairdensitywithheightfrom150to1000kmforhighandlowsolaractivity
and diurnal variations (see chapter 1.2) based on COSPAR International Reference Atmosphere
(CIRA1972). Courtesy: King-Hele[8]
5
Usingdimensionalanalysisfromfluiddynamics,wecandefinedragasfollows:
F =−1ρ(cid:175)(cid:175)v 2(cid:175)(cid:175)C A, (1.1)
d rel d
2
whereF isthedragforce;ρ isadensityofmedium,inourcase,thedensityofair;v isarelative
d rel
velocityofanobjectgoingthroughmedium;andAisacharacteristicareaofthatobject. C isadrag
d
coefficient; it has no units and depends on the object’s surface roughness and shape, as seen from
Figure1.2–italsodependsonthemedium’sdensityandthemolecularvelocity[9]. Nevertheless,
the dependence of different parameters for specific conditions are usually too small to take into
account,thusitismostlyusedasaconstant.
Whenmodelingaspaceobject,weusuallydonotknowitsshape,becausethisinformationisnot
savedinephemerides,asseenfromchapter2.1,hencewecannotknowwhatistheexactvalueofthe
dragcoefficient. Evenifwecouldcaptureanimageoftheobject’sprofileandmeasurethesurface
area,C and A valueswouldn’tberight,asmostoftheobjectsarenon-sphericalandspinningwith
d
respect to their direction of motion, hence one would need multiple pictures of different sides, as
wellas,toknowitsangularvelocity. Agoodwaytodealwiththisistoapproximatethemasspheres
and then use drag coefficient of a sphere. The value we use isC =2.2, which was popularized by
d
theuseofJacchiamodelsforestimatingatmosphericdensityandotherparametersofupperlayers
Figure1.2Dragcoefficientfordifferentshapeobjectscalculatedatlowvelocities[10]
6
of Earth’s atmosphere [11,12]. Note that this value is different from the one showed in Figure 1.2
asaresultofthefollowingassumptions[9]:
• Themeanfreepathoftheairmoleculesismuchgreaterthanthesizeofthesatellite.
• Thesatellitevelocityv ismuchgreaterthanthemeanrandomvelocityv¯oftheairmolecules.
0
• The mechanisms that deflect the impinging molecules range from elastic to inelastic impact
anddiffuseevaporationataneffectivesurfacetemperatureT.
1.2 Atmospheric Density Model
Asonecansee,thedragforcedependsontheatmosphericdensityindifferentways,butthedensity
is not constant. Even though we assume that its changes do little effect to the drag coefficient, for
thedragforcewehave touseaccuratevalues,asobjectsclosertothegroundwill interactwithita
lotmoreanddeorbitquickerthanthosewhoarefurtheraway;itiseasytoseethisfromtheISSand
Vanguard13 orbits. WhiletheISSisthebiggestman-madesatellite,ithasoneofthelowestorbits
of all artificial satellites and thus it loses 2 or more kilometers of altitude per month, as seen from
Figure 1.3. Without frequent orbital corrections it would deorbit within a year. Vanguard 1 is in a
higher orbit, with a perigee4 of 656 km and apogee5 of 3842 km (as of December 2015), and it is
expectedtostayinorbitfor240years6.
Figure 1.3 Orbital height of the ISS throughout last year. Image source: www.heavens-
above.com/ISSHeight.aspx
3Vanguard1istheoldestmanmadesatellitestillinorbit,launchedin1958;itwasusedtoobtaingeodeticmea-
surements.
4ApointwherecelestialbodyisclosesttotheEarth.
5ApointwherecelestialbodyisfurthestfromtheEarth.
6Retrievedfromhttp://nssdc.gsfc.nasa.gov/nmc/spacecraftDisplay.do?id=1958-002B.
7
The Earth’s atmosphere is expanding and contracting in a complex manner all the time, not to
mentionitisnotthesamethroughoutdifferentlatitudesandlongitudes. Manyeffectscontributeto
that and a good atmospheric model should include at least most significant ones, if not all. Below,
welisteffectsthatareconsideredtobemostimportant[13]:
• Latitudinal variations: The effect comes from satellite passing over the Earth’s equatorial
bulge,effectivelythischangestheactualaltitudeand,therefore,thedensity.
• Diurnal variations: These variations come from the Earth’s rotation and the Sun’s irradia-
tion. The day side of the Earth is warmer than the night side, hence the atmosphere expands
and bulges over ecliptic plane in the direction of the Sun. Because of the diurnal variations,
it’snecessarytoknowlatitude,localtime,andtimeoftheyear.
• 27-day solar-rotation cycle: This effect arises from Sun’s rotation around its axis, which
causesfluctuationsintheamountofradiationthatreachestheEarth.
• 11-year cycle of sunspots: Solar cycle strongly varies the amount of incoming radiation
reaching the Earth. As a result, the Earth’s atmosphere heats up different amounts and it
expands different amounts throughout the cycle. At the solar maximum the temperature of
upperlayersoftheatmosphereisabouttwotimeshigherthanatsolarminimum[14],meaning
that the atmosphere has expanded a lot more and debris will be affected more by the drag.
In fact, nine times more debris fall to the surface of the Earth during solar maximum than
minimum [15]. Furthermore, depending on the altitude, this effect can even cause larger
disturbancethroughsolar-radiationpressurethandrag[13].
• Seasonal variations: These variations come from solar flux fluctuations due to Earth’s ro-
tation around the Sun. The Earth’s rotational axis is tilted and its orbit is elliptical, thus the
Sun’sdeclinationandthedistancebetweentheEarthandtheSunvariesthroughouttheyear.
• Rotating atmosphere: The atmosphere is rotating together with the Earth, however, the
velocityoflowerlayerisusuallyhigherbecauseofthefrictionwiththeEarth’ssurface.
• Winds: Atmosphericwindsandweathercreatelocalvariationsofthetemperatureand,there-
fore, the density. Unfortunately, this is difficult to account for, as the weather itself is almost
unpredictablebecauseofitssensitivitytochanges.
• Magnetic-storm variations: Effect comes from fluctuations in the Earth’s magnetic field,
whichaffectstheamountofchargedparticlesreachingtheEarth’satmospherefromtheSun.
• Irregular short-periodic variations: Mostly solar flares: sudden increase in solar wind
intensity,whichalsodisturbstheEarth’smagneticfield.
8
• Tides: Ocean tides and even atmospheric tides as a result of gravitational pull of the Moon
cancausesmallvariationsintheatmosphericdensity.
The first model of the Earth’s atmosphere above 120 km was created by Nicolet in 1961 [16]. It
wasanempiricalmodelandusedasetofboundaryconditionsoftemperatureandpartialdensities
of gases altogether with diffusion theory to get values for higher altitudes. Later it was improved
uponseveraltimesbyJacchiawithempiricaldatafromthenewsatellites[11,17,18]. Itwasagood
model and even today it is still widely used for spacecraft dynamics. However, for more accurate
calculations like satellite and space debris orbital simulations, it is better to use a newer model,
developed by Mike Picone, Alan Hedin, and Doug Drobof, which is called NRLMSISE-00 [19]. It
isanabbreviationofUSNavalResearchLaboratoryandMassSpectrometerandIncoherentScatter
Radar (00 stands for the year of release), the two primary data sources for development of earlier
versionsofthemodel.
NRLMSISE-00 is an empirical model of the Earth’s atmosphere from the ground through ex-
osphere. Its outputs are atmospheric density, which we need, as well as temperature and other
parameters, while all the necessary input parameters are: year and day time of day; local apparent
solar time; geodetic altitude, latitude and longitude; daily and 81 day average of F solar flux7;
10.7
daily geomagnetic index. This model includes most of the effects listed above [19], it is imple-
mentedinLightForcecodeandthusitistheperfectmodelforoursimulations.
710.7cmsolarradioflux,F ,isagreatindicatorofsolaractivity,asitcorrelateswithsunspotsandUVradiation
10.7
[14],aswellasitiseasytomeasureitonthegroundusingradioantenna.
9
2 Numerical Approach
Fromthepreviouschapterwelearnedhowtheatmosphericdragaffectsthedebrisandwhyitvaries
in time, but the problem we face that it is very impractical to experimentally measure it without
knowing the shape and the size of the debris. However, we can find ratios between mass, drag
coefficient and characteristic surface area of debris – the ballistic coefficient, and it can be done
numericallyfromtheorbitaltrackingdata. Itisexplainedinthissection.
2.1 Input Data
Satellitetrackingbeganin1957withthelaunchofthefirstman-madesatelliteintospace–Sputnik.
Thefirstmethodsfortrackingusedradiocommunicationandweren’tveryreliableastheycouldn’t
detectortrackobjectsinspacethatweren’ttransmittinginternationallyagreedfrequencies,assome
did. At that time, the main reason for satellite tracking was to defend against missile attacks, as a
resultbettermethodswereconceived. Telescopeswithcameraswereusedinstead,aswellasradars,
although they had their own limitations. Optical tracking systems depended on the weather and
only worked during night time, while radar was free from those problems, but it was costly [20].
Nowadaysthemethodsarethesame,onlytheaccuracyhasincreasedabit,albeitstillrelativelylow,
andobjectsassmallas1cmcouldnowbedetectedandtrackedwithladars[21],whichisgreatfor
estimatingvariousparametersoftheatmosphere,aswellaspredictingtrajectories.
Themostpopularwayforacelestialbody’sdetectionandtrackingisusingaradarsystem. The
detection is done by sending a powerful, yet short, impulse of radio signal into particular part of
space and waiting for echoes. Computer analyzes the returned signal and from its strength, shape
andtimeintooktocomebackitcalculatesaverageephemerisforthatobject[22]. Tracking,onthe
other hand, uses a priori knowledge where the object is going to be at specific time, so the radar
is prepared by pointing to that part of the sky. After detection, it then tracks that object to get new
orbital data, which is then saved as one data set. But there is only a finite number of radar sensor
sites around the world, and they have a narrow view of the sky, and limited capabilities, meaning
thatsomespaceobjectaren’tinviewallthetime. Furthermore,theymightnotbeinagoodviewor
trackedforcoupledays,thusmeasurementsarerare–frompublicdatabase8 weseethatitisabout
onepertwodaysorless.
Theformatforstoringephemeridesisthesameasitwas50yearsago. Backthen,asetoforbital
elements were written in two lines on a punch card9; from here came the name Two-Line Element
(TLE) set, meaning this data encoding format. Given a reference frame and an epoch (a reference
time when ephemeriswere taken), actually onlysix orbital elements areneeded to unambiguously
8www.celestrak.com/NORAD/elements/
9Punchcardisapieceofstiffpaperthatcontainedeithercommandsforcontrollingautomatedmachineryordata
for data processing applications. Both commands and data were represented by the presence or absence of holes in
predefinedpositions.
10